Server Disaggregation Is Making Cloud Data Centers Work Efficiently
With the rise of cloud computing, the spotlight is now on data centers. Together they consume around seven percent of the total global electricity supply, and the percentage is growing with every tick of the clock. The hunt, therefore, is now on for means that can make IT infrastructure ever more energy efficient. Companies are now putting more efforts to transform the way computers and data centers are built.
At present, IT infrastructure’s basic building blocks are the servers. But the problem is that there is a limit to which the memory of CPU resources can be utilized. Also, servers cannot easily share any spare resources outside of their own box. So the IT Industry is trying to compose servers into software control solely to have as many CPUs and memory as per the requirement of any particular workload. Disaggregated computer architecture is incidentally a hot topic; it basically separates servers into their constituent compute and memory resources. This allows the memory resources to be allocated according to the needs of different workloads.
The dReDBox project is a vital stepping stone in that direction. This project will use discrete compute, and memory modules known as bricks-connected together by high-speed links-which will enable a server to be framed for a specific application. This can be done with as many CPUs cores and memory, depending on the job to be executed.